T cell–mediated immune response plays a crucial role in controlling Trypanosoma cruzi infection and parasite burden, but it is also involved in the clinical onset and progression of chronic Chagas’ disease. Therefore, the study of T cells is central to the understanding of the immune response against the parasite and its implications for the infected organism. The complexity of the parasite–host interactions hampers the identification and characterization of T cell–activating epitopes. We approached this issue by combining in silico and in vitro methods to interrogate patients’ T cells specificity. Fifty T. cruzi peptides predicted to bind a broad range of class I and II HLA molecules were selected for in vitro screening against PBMC samples from a cohort of chronic Chagas’ disease patients, using IFN-γ secretion as a readout. Seven of these peptides were shown to activate this type of T cell response, and four out of these contain class I and II epitopes that, to our knowledge, are first described in this study. The remaining three contain sequences that had been previously demonstrated to induce CD8+ T cell response in Chagas’ disease patients, or bind HLA-A*02:01, but are, in this study, demonstrated to engage CD4+ T cells. We also assessed the degree of differentiation of activated T cells and looked into the HLA variants that might restrict the recognition of these peptides in the context of human T. cruzi infection.

In the last decades, the importance of T cells in adaptive immunity against chronic infection with parasite Trypanosoma cruzi has been profusely demonstrated. T. cruzi–specific T cells are crucial for the maintenance of the low or undetectable numbers of circulating parasites typically seen in the chronic infection (1), and they seem to make an important contribution to the absence of clinical manifestations in asymptomatic chronic patients (2, 3). However, T cells are also involved in inflammatory imbalance, tissue damage, and the cardiomyopathy associated with this infection (410).

Several T. cruzi epitopes have been proven to generate specific memory T cells, detectable (albeit in low frequencies) in the circulation of chronically infected patients. In particular, cruzipain, kinetoplastid membrane protein 11 (KMP-11), heat shock protein 70 (HSP-70), putative surface Ag TcCA-2, and several members of the trans-sialidase superfamily of proteins have been reported as sources of epitopes recognized by human CD8+ T cells (1115). The specificity of T. cruzi–specific CD4+ T cells has been studied to a far lesser extent (16), and only a few proteins (trypomastigote surface Ag 1 [TSA-1], Tc24, Ag B13, and KMP-11) have been demonstrated to induce a recall response on CD4+ T cells cells from chronic patients (1719).

The characterization of Ag-specific human T cells is constrained by their low frequency in circulation (20). This is particularly true for the case of chronic, latent infections, as opposed to active infections that produce an expansion and mobilization of pathogen-specific T cells, increasing their frequency in peripheral blood. Working with protozoan parasitic infections often entails additional obstacles, as these organisms usually have large genomes and complex proteomes: the genome of T. cruzi has the potential to encode over 1.2 × 104 proteins (21, 22), which, redundancy aside, may produce several hundreds of thousands of different HLA-binding peptides. Because, in most settings, it is unfeasible to exhaustively test all of these for a certain pathogen, discriminating which HLA-binding peptides have immune relevance for protection and pathogenesis remains a challenging issue. Given this, a strategy to narrow down the possibilities to a scale that enables experimental testing becomes indispensable. Possible workaround schemes for this issue include prediction of peptide binding to a representative set of HLA alleles and analysis of a selected, limited subset of the pathogen proteome (23).

In this study, we applied a combination of these two, by using an immunoinformatic predictive strategy to identify peptides from 53 T. cruzi proteins with high HLA-I– and HLA-II–binding potential, to then assess their capability to induce a recall T cell response in samples from chronically infected human subjects. We describe seven novel parasite-derived peptides that are recognized, to different extents, by CD4+ and CD8+ T cells. Out of these, three contain previously known class I epitopes but are, in this study, demonstrated to activate CD4+ T cells.

Chronic Chagas’ disease adult patients were recruited at Instituto Nacional de Parasitología Dr. M. Fatala Chabén and Hospital General de Agudos Dr. I. Pirovano. The sampling process followed the tenets of the Declaration of Helsinki and of both institutions’ medical ethics committees. In accordance, all included subjects gave written informed consent prior to sample extraction and after the nature of this study was explained to them.

Included subjects had a minimum of two positive serological tests for Chagas’ disease and lived within the greater Buenos Aires area (nonendemic for the infection) at the time of blood collection. A total of 51 chronic Chagas’ disease patients, for whom clinical classification and epidemiological data are summarized in Supplemental Table I, were recruited and classified as asymptomatic (without demonstrable cardiomyopathy; n = 25) or cardiac (n = 26). Nine noninfected subjects were included as controls.

Samples consisted of 35–50 ml peripheral venous blood, collected in EDTA anticoagulated tubes and processed up to 4 h after collection. PBMCs were isolated by centrifugation (400 × g at room temperature for 40 min) in a Ficoll-Paque gradient medium (GE Healthcare Bio-Sciences, Uppsala, Sweden), quantified by manual count in Neubauer chamber, and aliquoted in FBS (Natocor, Córdoba, Argentina) with 10% v/v DMSO to be cryopreserved in liquid nitrogen until used. Total DNA was extracted from whole-blood aliquots using the High Pure PCR Template Preparation Kit (Roche Diagnostics, Indianapolis, IN) following manufacturer-provided instructions. Plasma was obtained by whole-blood centrifugation at 400 × g at room temperature for 10 min.

The sequences of all T. cruzi proteins listed as T cell immunogens at the Immune Epitope Database (http://www.iedb.org/) (24) were screened in silico to extract all possible 15-aa-long sequences conserved between Sylvio X10 and CL Brener proteomes, as sourced from the TriTryp database (http://www.tritrypdb.org, version 6.0) (25). At the time the search was performed, the list included 53 distinct proteins and yielded a total of 4387 unique 15-mers after filtering out all sequences with a 9-aa overlap with the human proteome.

For this subset of peptides, all unique 9-mer and 10-mer peptides were extracted, and binding to a set of 35 HLA-A and HLA-B molecules prevalent in Latin America (Supplemental Fig. 1) was predicted using the NetMHCpan (26) method (version 2.8). Peptides were defined as binders if their predicted binding affinity was stronger than 50 nM or if their predicted percentile ligand likelihood rank score was ≤0.5%. Likewise, binding was predicted for each 15-mer peptide to a set of 11 HLA-DRB1 molecules, selected based on worldwide prevalence (Supplemental Fig. 1), using NetMCIIpan (27) (version 2.1). In this study, binding peptides were defined as those with a predicted rank score ≤5%.

Given this set of 15-mers mapped to each of the included HLA class I and class II molecules, the PopCover method (28) was used to select a set of 15-mer peptides that each had a high degree of HLA promiscuity and, in concert, gave the broadest coverage of the different HLA molecules and the aforementioned 53 T. cruzi Ags. Finally, in situations in which two 15-mer peptides sharing an overlap of 14 aa were selected, these two peptides were merged into one 16-mer peptide.

Whole antigenic lysate from T. cruzi epimastigote was prepared from axenic cultures (CL Brener strain) in liver infusion tryptose medium, as previously described (29). After lysis, the suspension was filter sterilized through a 0.2-μm-pore-size membrane, aliquoted, and stored at −80°C until use.

Peptides were synthesized by GenScript (Hong Kong Island, Hong Kong SAR) and resuspended in DMSO to a final concentration of 10 μg/μl. The 50 peptides were randomly distributed in five groups of 10 peptides each to form peptide pools I to V. To prepare these, equal volumes of each peptide solution were mixed, yielding a pooled solution of 10 μg/μl total peptides (1 μg/μl of each peptide). DMSO was added in the remaining experimental condition wells to control its potential effect on cellular response. PHA-A (PHA) was used at a concentration of 5 μg/ml, as nonspecific, positive-control condition.

The minimal epitope TcCZp2me was predicted as the top-scoring core HLA-A*31:01–binding peptide embedded in the full-length TcCZp2, using NetMHCpan (version 3.0). The synthetic peptide was purchased from ImmunAware (Copenhagen, Denmark).

The Human IFN-γ ELISPOT Set and the AEC ELISPOT Substrate Set (BD Biosciences, San Diego, CA) were used to estimate the frequency of IFN-γ–producing, epitope-specific T cells following manufacturer-provided instructions. For each replicate, 4 × 105 PBMC (or equivalent adjusted number of CD4- or CD8-depleted PBMC, see below) per well were seeded in supplemented RPMI medium (RPMI 1640 medium, 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM l-glutamine, and 10% heat-inactivated FBS). In this setup, cells were stimulated for 18–20 h with peptide pools (10 μg/ml [i.e., 1 μg/ml of each peptide]), individual peptides (10 μg/ml), T. cruzi lysate (10 μg/ml), or PHA (5 μg/ml).

Response is depicted either as raw spot-forming unit (SFU) counts or as logarithmic stimulation index, calculated as the base 2 logarithm of the ratio between the mean SFU values in the challenge condition and the nonstimulated control condition. This transformation allows the symmetric representation of positive and negative fold changes in the SFU values.

The EasySep Human CD4 or CD8 Positive Selection Kit and the EasySep Purple Magnet (STEMCELL Technologies, Vancouver, Canada) were used to deplete these populations from whole PBMC following the manufacturer-provided protocol. The CD3+CD4+ and CD3+CD8+ lymphocyte frequencies prior and after magnetic separation were assessed by flow cytometry on a BD FACSCanto II cytometer (BD Biosciences) and used to estimate the remaining proportion of the depleted population in each case. CD4+ and CD8+ T cells were subtracted to an extent of 81.4 ± 14.9% and 98.1 ± 0.3%, respectively. For the IFN-γ ELISPOT experiments performed with these population-depleted PBMC, the number of cells to seed per well were adjusted for each sample on the basis of cell counts before and after depletion.

The HLA-A, HLA-B, HLA-DPA, HLA-DPB, HLA-DQA, HLA-DQB, HLA-DRB1, and HLA-DRB3 loci were genotyped for each sample by next-generation sequencing at the Institute of Clinical Molecular Biology at Christian-Albrechts-University of Kiel (Kiel, Germany) by a method described elsewhere (30).

FITC anti-CD3, PE-Cy5 anti-CD8, allophycocyanin anti-CD45RA, and PE anti-CD45RO (BD Biosciences) and PE-Cy7 anti-CD4 and allophycocyanin–Cy7 anti-CCR7 (BioLegend, San Diego, CA) Abs were used to stain cell surface phenotype markers. Zombie Aqua Fixable Dye (BioLegend) was used to gate out nonviable cell events.

BV421-labeled peptide–HLA multimeric complexes were purchased from ImmunAware. Cells were thawed, washed in PBS, and immediately submitted to the staining protocol detailed below or incubated in supplemented RPMI medium for 7 d (31) in the presence or absence of 10 μg/ml peptide (either TcCZp2 or TcCZp2me) prior to multimer staining. In accordance with manufacturer instructions, cells were first stained with peptide–HLA multimers (20 min at room temperature in the case of class I multimers and 1 h at 37°C for class II multimers) diluted in supplemented RPMI medium, washed with cold PBS 1% FBS, stained with anti–surface marker Abs, and washed again prior to fixation.

For intracellular cytokine staining experiments, 5 × 105 PBMC were seeded in 96-well, U-bottom plates and incubated for 18 h under the same condition as used for the IFN-γ ELISPOT experiments, with the addition of BD FastImmune CD28/CD49d Costimulatory Ab mixture (BD Biosciences). Brefeldin-A (5 μg/ml) and monensin (2 nM; BioLegend) were added for the last 4 h of this incubation. Cells were washed and stained for surface markers as detailed above, then permeabilized with 1× Perm Buffer (BioLegend) and stained with BV421 anti–IFN-γ Ab (BioLegend). Two washing steps with 1× Perm Buffer were performed before fixation.

In all cases, samples were fixated with 1× Fixation Buffer (BioLegend), washed, and resuspended in PBS. Experiments were acquired in a BD FACSCanto II flow cytometer (BD Biosciences) and analyzed using a customized pipeline based on R package openCyto (32) and the directed automated filtering and identification of cell populations (DAFi) gating method (33). Briefly, the openCyto framework was used to automate gating from different experiments in a semisupervised manner, and the resulting gates were translated into rectangular coordinates that were used as input for the DAFi gating strategy. Plots were generated using ggplot2 (34). The R code for this implementation is available upon request.

ELISPOT results were analyzed by distribution-free resampling (35) implemented in R.

To identify candidate epitopes, we designed and performed a predictive in silico approach that we used to rank a set of 4387 T. cruzi peptides based on predicted HLA-binding strength and promiscuity (Fig. 1A). The top-scoring 50 peptides (Supplemental Table II) selected for in vitro screening were randomized in five equal-sized pools (identified with Roman numerals I–V) and used to challenge PBMC from a cohort of 30 chronic Chagas’ disease patients (15 asymptomatic and 15 cardiac subjects). Peptide-specific and whole-parasite lysate responses were evaluated as IFN-γ secretion by ELISPOT (Fig. 1B). Chagas’ disease patients were segregated into responders and nonresponders according to their reactivity against T. cruzi lysate, whereas all the control subjects were nonresponders. Peptide pool-specific responses were observed in several patients, with different intensities, implying that one or more peptides within the positive pools were able to trigger IFN-γ secretion. Remarkably, patients responding to one or more of the peptide pools also responded to the lysate, regardless of their asymptomatic or cardiac clinical status, which suggests an association of peptide reactivity to a measurable memory response against T. cruzi Ags.

FIGURE 1.

In silico prediction and in vitro evaluation of T. cruzi T cell epitopes. (A) Schematic representation of the bioinformatic pipeline used for prediction. (B) IFN-γ response against T. cruzi lysate (Tc) and T. cruzi peptide pools (I–V) measured by ELISPOT on PBMC from chronic Chagas’ disease patients and non–T. cruzi–infected control subjects. The logarithmic stimulation index (LSI) was calculated as described in methods. LSI = 0 (solid line) means equal responses in control (unstimulated [U]) and stimulated conditions, whereas LSI = 1 when SFU in the stimulated condition is twice as many as in the control condition. Patients were grouped depending on whether they presented (Resp.) or not (Non resp.) a measurable response against parasite lysate. (C) Deconvolution ELISPOT for peptide pools I, II, III, and IV, using available PBMC samples from patients with positive response in the experiment from (B). Pool V was not deconvoluted because of the response observed against it being, although statistically significant, of low intensity. Each point represents the SFU count for an individual replicate well. Stimuli are indicated on the x-axis. Roman numerals indicate peptide pools. Numbers correspond to individual peptide identifiers (IDs), as detailed on Supplemental Table II. Asterisks indicate statistically significant difference versus the U condition. IDs beginning with A and C indicate donors classified as asymptomatic and cardiac patients, respectively. *p < 0.05, distribution-free resampling test.

FIGURE 1.

In silico prediction and in vitro evaluation of T. cruzi T cell epitopes. (A) Schematic representation of the bioinformatic pipeline used for prediction. (B) IFN-γ response against T. cruzi lysate (Tc) and T. cruzi peptide pools (I–V) measured by ELISPOT on PBMC from chronic Chagas’ disease patients and non–T. cruzi–infected control subjects. The logarithmic stimulation index (LSI) was calculated as described in methods. LSI = 0 (solid line) means equal responses in control (unstimulated [U]) and stimulated conditions, whereas LSI = 1 when SFU in the stimulated condition is twice as many as in the control condition. Patients were grouped depending on whether they presented (Resp.) or not (Non resp.) a measurable response against parasite lysate. (C) Deconvolution ELISPOT for peptide pools I, II, III, and IV, using available PBMC samples from patients with positive response in the experiment from (B). Pool V was not deconvoluted because of the response observed against it being, although statistically significant, of low intensity. Each point represents the SFU count for an individual replicate well. Stimuli are indicated on the x-axis. Roman numerals indicate peptide pools. Numbers correspond to individual peptide identifiers (IDs), as detailed on Supplemental Table II. Asterisks indicate statistically significant difference versus the U condition. IDs beginning with A and C indicate donors classified as asymptomatic and cardiac patients, respectively. *p < 0.05, distribution-free resampling test.

Close modal

Moving forward, we sought to narrow down the identity of T cell activators to individual peptides within the pools. Thus, positive patient-peptide pool pairs were selected based on response and sample availability and assayed again, this time using individual peptides to deconvolute the IFN-γ response induced against the pool (Fig. 1C). Seven individual peptides were found to activate IFN-γ secretion in at least one of the patients assayed. Three of these contain epitopes from cruzipain and actual/putative trans-sialidase(s), which had been previously described (13, 14, 3638). The remaining four are or comprise novel T. cruzi T cell epitopes, also from cruzipain and putative trans-sialidase sequences (Table I). For ease of interpretation, these seven peptides were assigned identifiers, as detailed in the same table.

Table I.
Peptides found to induce IFN-γ secretion from PBMC isolated from chronic Chagas’ diseasepatients
PeptideIDSequenceAgaPrior Evidence
TcTSp1 ENQLYHFANYKFTLV Trans-sialidase, group V, putative Contains class I epitopes TSKB20 (ANYKFTLV) and TSA2-44 (FANYKFTLV), reported as immunogenic in mice and humans, respectively (20, 41, 43,74). In vivo CD8+ T cell–mediated cytotoxicity and ex vivo IFN-γ release were observed in cells from mice immunized with TSKB20 (41) or infected with T. cruzi (42); TSKB20-specific cells were found in chronically infected and benznidazole-treated-infected mice (42, 43). Ex vivo IFN-γ secretion was observed by ELISPOT using cells from HLA-A*02:01+ human subjects with chronic Chagas disease, challenged with TSA2-44 (74). Mice H2kb and human HLA-A MHC molecules binding was demonstrated in vitro. 
10 TcTSp2 TVPYHFANSKFTLVA Flagellum-associated protein FL-160-2 Contains epitopes FANSKFTLV (19), ANSKFTLV (20), and FANSKFTLVA (19), which were found by mouse (H-2kb) and human (HLA-A*02) MHC-binding prediction but resulted negative in IFN-γ ELISPOT experiments in both species. 
15 TcCZp1 VECQWFLAGHPLTNLS Cruzipain/papain family cysteine protease/major cysteine protease Contains epitope FLAGHPLTNL, which was demonstrated to bind the MHC molecule encoded by HLA-A*02:01 genes in vitro, but which failed to induce IFN-γ secretion on patient-derived cells in ELISPOT experiments (19). 
16 TcTSp3 MLSLVAAVKAPRTHN Trans-sialidase (dm28c)/trans-sialidase, putative, group V (CL Brener, Esmeraldo-like), groups II, V, and VI (CL Brener, non–Esmeraldo-like) Novel 
20 TcTSp4 GVVMEDGTLVFPLMA Trans-sialidase (CL Brener)/trans-sialidase, putative/trans-sialidase (pseudogenic product)/gp82/BNR repeats–like domain Novel 
38 TcTsp5 HRFTLVATVTIHQVPK ASP-2/trans-sialidase/trans-sialidase group II, putative/trans-sialidase (pseudogenic product)/BNR repeat–like domain Novel 
47 TcCZp2 RQRRYQPYHSRHRRL Cruzipain/cysteine peptidase, putative/protein of unknown function (Sylvio)/major cysteine protease (CL Brener, non–Esmeraldo-like) Novel 
PeptideIDSequenceAgaPrior Evidence
TcTSp1 ENQLYHFANYKFTLV Trans-sialidase, group V, putative Contains class I epitopes TSKB20 (ANYKFTLV) and TSA2-44 (FANYKFTLV), reported as immunogenic in mice and humans, respectively (20, 41, 43,74). In vivo CD8+ T cell–mediated cytotoxicity and ex vivo IFN-γ release were observed in cells from mice immunized with TSKB20 (41) or infected with T. cruzi (42); TSKB20-specific cells were found in chronically infected and benznidazole-treated-infected mice (42, 43). Ex vivo IFN-γ secretion was observed by ELISPOT using cells from HLA-A*02:01+ human subjects with chronic Chagas disease, challenged with TSA2-44 (74). Mice H2kb and human HLA-A MHC molecules binding was demonstrated in vitro. 
10 TcTSp2 TVPYHFANSKFTLVA Flagellum-associated protein FL-160-2 Contains epitopes FANSKFTLV (19), ANSKFTLV (20), and FANSKFTLVA (19), which were found by mouse (H-2kb) and human (HLA-A*02) MHC-binding prediction but resulted negative in IFN-γ ELISPOT experiments in both species. 
15 TcCZp1 VECQWFLAGHPLTNLS Cruzipain/papain family cysteine protease/major cysteine protease Contains epitope FLAGHPLTNL, which was demonstrated to bind the MHC molecule encoded by HLA-A*02:01 genes in vitro, but which failed to induce IFN-γ secretion on patient-derived cells in ELISPOT experiments (19). 
16 TcTSp3 MLSLVAAVKAPRTHN Trans-sialidase (dm28c)/trans-sialidase, putative, group V (CL Brener, Esmeraldo-like), groups II, V, and VI (CL Brener, non–Esmeraldo-like) Novel 
20 TcTSp4 GVVMEDGTLVFPLMA Trans-sialidase (CL Brener)/trans-sialidase, putative/trans-sialidase (pseudogenic product)/gp82/BNR repeats–like domain Novel 
38 TcTsp5 HRFTLVATVTIHQVPK ASP-2/trans-sialidase/trans-sialidase group II, putative/trans-sialidase (pseudogenic product)/BNR repeat–like domain Novel 
47 TcCZp2 RQRRYQPYHSRHRRL Cruzipain/cysteine peptidase, putative/protein of unknown function (Sylvio)/major cysteine protease (CL Brener, non–Esmeraldo-like) Novel 
a

Ag information was sourced from TriTrypDB.

ID, identifier.

To estimate the frequency of responding patients within the population, each of the seven peptides identified in the pool deconvolution experiments was next used to challenge an extended set of patient samples (n = 51, 25 asymptomatic and 26 cardiac Chagas’ disease patients) in IFN-γ ELISPOT experiments. The results in Table II present the percentages of patient samples with a detectable response against these peptides. As the data show, 37.3% of the samples in the cohort responded to at least one of the peptides. Differences in the percentage of responding subjects between the asymptomatic and cardiac chronic Chagas groups of patients were statistically nonsignificant (Fisher exact test, p = 0.39).

Table II.
Frequency (expressed as counts and percentage) of responding patients according to the number of peptides they responded to in IFN-γ ELISPOT experiments
ResponseAsymptomatic (n = 25)Cardiac (n = 26)Overall (n = 51)
Negative 14/25 (56.0%) 18/26 (69.2%) 32/51 (62.7%) 
Positive 11/25 (44.0%) 8/26 (30.8%) 19/51 (37.3%) 
 One peptide 8/25 (32.0%) 7/26 (26.9%) 15/51 (29.4%) 
 Two peptides 3/25 (12.0%) 0/26 (0.0%) 3/51 (5.9%) 
 Four peptides 0/25 (0.0%) 1/26 (3.8%) 1/51 (2.0%) 
ResponseAsymptomatic (n = 25)Cardiac (n = 26)Overall (n = 51)
Negative 14/25 (56.0%) 18/26 (69.2%) 32/51 (62.7%) 
Positive 11/25 (44.0%) 8/26 (30.8%) 19/51 (37.3%) 
 One peptide 8/25 (32.0%) 7/26 (26.9%) 15/51 (29.4%) 
 Two peptides 3/25 (12.0%) 0/26 (0.0%) 3/51 (5.9%) 
 Four peptides 0/25 (0.0%) 1/26 (3.8%) 1/51 (2.0%) 

When the breadth of the response was analyzed, in terms of the number of peptides recognized by each patient, results indicated that most of the responding subjects produced a significant IFN-γ secretion against only one peptide (Table II). In addition, the peptides most frequently recognized by T cells were TcTSp4 (GVVMEDGTLVFPLMA; 13.7% responsive patients) and TcTSp5 (HRFTLVATVTIHQVPK; 19.6% responsive patients; Table III). For each individual peptide, no differences between the asymptomatic and cardiac groups were observed in the frequency of responding patients (Fisher exact test, p > 0.25). In summary, the observed response is, in most patients, restricted to only one of the analyzed peptides, or none, and it shows no apparent relation with their degree of cardiac compromise.

Table III.
Frequency (expressed as counts and percentage) of patients responding to each peptide within the cohort
Responding Patients
PeptideAsymptomatic (n = 25)Cardiac (n = 26)Overall (n = 51)
TcTSp1 2/25 (8.0%) 1/26 (3.8%) 3/51 (5.9%) 
TcTSp2 1/25 (4.0%) 1/26 (3.8%) 2/51 (3.9%) 
TcTsp3 0/25 (0.0%) 2/26 (7.7%) 2/51 (3.9%) 
TcTSp4 5/25 (20.0%) 2/26 (7.7%) 7/51 (13.7%) 
TcTSp5 5/25 (20.0%) 5/26 (19.2%) 10/51 (19.6%) 
TcCZp1 0/25 (0.0%) 1/26 (3.8%) 1/51 (2.0%) 
TcCZp2 2/25 (8.0%) 1/26 (3.8%) 3/51 (5.9%) 
Responding Patients
PeptideAsymptomatic (n = 25)Cardiac (n = 26)Overall (n = 51)
TcTSp1 2/25 (8.0%) 1/26 (3.8%) 3/51 (5.9%) 
TcTSp2 1/25 (4.0%) 1/26 (3.8%) 2/51 (3.9%) 
TcTsp3 0/25 (0.0%) 2/26 (7.7%) 2/51 (3.9%) 
TcTSp4 5/25 (20.0%) 2/26 (7.7%) 7/51 (13.7%) 
TcTSp5 5/25 (20.0%) 5/26 (19.2%) 10/51 (19.6%) 
TcCZp1 0/25 (0.0%) 1/26 (3.8%) 1/51 (2.0%) 
TcCZp2 2/25 (8.0%) 1/26 (3.8%) 3/51 (5.9%) 

Next, we set ourselves to determine whether the observed IFN-γ secretion was associated to the activation of CD4+ T cells, CD8+ T cells, or both. IFN-γ ELISPOT experiments were performed using CD4+ or CD8+ T cell–depleted PBMC. As depicted in Fig. 2, CD4+ T cells were the main producers of IFN-γ in response to peptides TcTSp1, TcTSp2, TcTSp3, TcTSp4, TcTSp5, and TcCZp1. In contrast, only TcCZp2 showed a predominant CD8+ T cell activation.

FIGURE 2.

IFN-γ response against T. cruzi peptides from CD4-depleted (filled arrowheads) or CD8-depleted (hollow arrowheads) PBMC. SFU measurements in each depletion condition were normalized to the values observed for nondepleted PBMC from the same patient (vertical axes). Horizontal axes show the estimated frequency of remaining CD4+ (top row) or CD8+ (bottom row) T cells after depletion. Patient identifiers (IDs) are indicated at the end of each line. In the cases of peptides with <3 responsive patients (i.e., TcCZp1, TcTSp1, TcTSp2, and TcTSp3), data were grouped in a single plot (Other peptides). In this case, the peptide is specified next to the corresponding patient ID. Donor IDs beginning with A and C indicate donors classified as asymptomatic and cardiac patients, respectively.

FIGURE 2.

IFN-γ response against T. cruzi peptides from CD4-depleted (filled arrowheads) or CD8-depleted (hollow arrowheads) PBMC. SFU measurements in each depletion condition were normalized to the values observed for nondepleted PBMC from the same patient (vertical axes). Horizontal axes show the estimated frequency of remaining CD4+ (top row) or CD8+ (bottom row) T cells after depletion. Patient identifiers (IDs) are indicated at the end of each line. In the cases of peptides with <3 responsive patients (i.e., TcCZp1, TcTSp1, TcTSp2, and TcTSp3), data were grouped in a single plot (Other peptides). In this case, the peptide is specified next to the corresponding patient ID. Donor IDs beginning with A and C indicate donors classified as asymptomatic and cardiac patients, respectively.

Close modal

In the light of these results and those from the response breadth analysis, we decided to extend our exploration of HLA restriction. Full-resolution HLA typing (30) was performed for each of the patients in the cohort. A comparison of the resulting list of alleles and the set of alleles used for the epitope prediction is shown in Supplemental Fig. 1A. By identifying HLA variants (class I or II HLA, depending on whether CD8+ or CD4+ predominance was observed in IFN-γ secretion) shared by patients who responded to a given peptide, we sought to identify the most likely HLA restriction element for our epitopes. In parallel, NetMHCpan (version 4.0) and NetMHCIIpan (version 3.2) were used to predict binding of every HLA molecule expressed in the patients in the cohort to each of the seven candidate peptides. The combination of these results for peptides with at least three responding patients allowed us to postulate candidate HLA restriction elements for each epitope. The resulting inference analysis is summarized in Table IV and Supplemental Table III, and its main conclusions are outlined below.

Table IV.
Inference analysis
Peptide IDCore Peptide SequenceaHLA AlleleAffinity (nM)Rank (%)Allele freq. in Responding Patients
Allele freq. in Nonresponding Patients
TcTSp1 
 ENQLYHFANYKFTLV DPA1*01:03-DPB1*04:02 283.59 3.00 100.0% (3/3) 54.20% (26/48) 
TcTSp4 
 GVVMEDGTLVFPLMA DRB3*01:01 68.31 1.30 33.3% (2/6) 33.30% (15/45) 
 DQA1*01:01-DQB1*03:02 2294.31 1.80 33.3% (2/6) 2.20% (1/45) 
 DQA1*03:01-DQB1*03:02 1037.12 4.00 33.3% (2/6) 53.30% (24/45) 
 DQA1*01:01-DQB1*06:02 838.52 4.50 16.7% (1/6) 0.00% (0/45) 
TcTSp5 
 HRFTLVATVTIHQVP DRB1*07:01 36.14 1.70 57.1% (4/7) 11.40% (5/44) 
 DRB1*08:02 150.96 2.00 28.6% (2/7) 29.50% (13/44) 
 DRB1*04:07 352.30 2.50 28.6% (2/7) 20.50% (9/44) 
 DQA1*02:01-DQB1*03:02 417.39 2.50 42.9% (3/7) 2.30% (1/44) 
 DPA1*01:03-DPB1*03:01 413.33 2.50 14.3% (1/7) 15.90% (7/44) 
 DPA1*01:03-DPB1*104:01 413.33 2.50 14.3% (1/7) 0.00% (0/44) 
 DQA1*02:01-DQB1*06:02 208.87 3.50 14.3% (1/7) 0.00% (0/44) 
 DRB1*04:05 116.43 3.50 14.3% (1/7) 0.00% (0/44) 
 DQA1*01:02-DQB1*06:02 246.53 4.50 14.3% (1/7) 13.60% (6/44) 
 DRB1*16:02 100.73 4.50 28.60% (2/7) 2.30% (1/44) 
 DPA1*01:03-DPB1*11:01 576.56 4.50 14.30% (1/7) 0.00% (0/44) 
TcCZp2 
 RYQPYHSRHR A*31:01 19.20 0.14 100.00% (3/3) 20.80% (10/48) 
 RQRRYQPYHSR 32.05 0.25 
 RYQPYHSRHRR 38.68 0.32 
Peptide IDCore Peptide SequenceaHLA AlleleAffinity (nM)Rank (%)Allele freq. in Responding Patients
Allele freq. in Nonresponding Patients
TcTSp1 
 ENQLYHFANYKFTLV DPA1*01:03-DPB1*04:02 283.59 3.00 100.0% (3/3) 54.20% (26/48) 
TcTSp4 
 GVVMEDGTLVFPLMA DRB3*01:01 68.31 1.30 33.3% (2/6) 33.30% (15/45) 
 DQA1*01:01-DQB1*03:02 2294.31 1.80 33.3% (2/6) 2.20% (1/45) 
 DQA1*03:01-DQB1*03:02 1037.12 4.00 33.3% (2/6) 53.30% (24/45) 
 DQA1*01:01-DQB1*06:02 838.52 4.50 16.7% (1/6) 0.00% (0/45) 
TcTSp5 
 HRFTLVATVTIHQVP DRB1*07:01 36.14 1.70 57.1% (4/7) 11.40% (5/44) 
 DRB1*08:02 150.96 2.00 28.6% (2/7) 29.50% (13/44) 
 DRB1*04:07 352.30 2.50 28.6% (2/7) 20.50% (9/44) 
 DQA1*02:01-DQB1*03:02 417.39 2.50 42.9% (3/7) 2.30% (1/44) 
 DPA1*01:03-DPB1*03:01 413.33 2.50 14.3% (1/7) 15.90% (7/44) 
 DPA1*01:03-DPB1*104:01 413.33 2.50 14.3% (1/7) 0.00% (0/44) 
 DQA1*02:01-DQB1*06:02 208.87 3.50 14.3% (1/7) 0.00% (0/44) 
 DRB1*04:05 116.43 3.50 14.3% (1/7) 0.00% (0/44) 
 DQA1*01:02-DQB1*06:02 246.53 4.50 14.3% (1/7) 13.60% (6/44) 
 DRB1*16:02 100.73 4.50 28.60% (2/7) 2.30% (1/44) 
 DPA1*01:03-DPB1*11:01 576.56 4.50 14.30% (1/7) 0.00% (0/44) 
TcCZp2 
 RYQPYHSRHR A*31:01 19.20 0.14 100.00% (3/3) 20.80% (10/48) 
 RQRRYQPYHSR 32.05 0.25 
 RYQPYHSRHRR 38.68 0.32 
a

IFN-γ ELISPOT, HLA typing, and in silico prediction data were combined to identify probable restriction elements for peptides TcTSp1, TcTSp4, TcTSp5, and TcCZp2. Rows were sorted based on Rank. Only the peptides predicted to bind an HLA molecule of their respective class (i.e., Rank < 5% for class II HLA alleles and Rank < 0.5% for class I HLA alleles) are shown. The complete, nonfiltered data table can be accessed as supplementary information (Supplemental Table III).

freq., frequency; ID, identifier; Rank, prediction percentile ligand likelihood rank.

The combination HLA-DPA1*01:03-DPB1*04:02 was the only HLA molecule predicted as strong binder to peptide TcTSp1 present in all responding patients. Note, however, that this allele combination was also frequent (45.8%) among the nonresponding patients.

All except one of the patients with a specific IFN-γ response against peptide TcTSp4 had at least one class II HLA molecule predicted to strongly bind this peptide: HLA-DQA1*03:01-DQB1*03:02, -DQA1*01:01-DQB1*03:02, -DQA1*01:01-DQB1*06:02, and -DRB3*01:01. However, these alleles had similar occurrence among patients with response against this peptide, and therefore, a most likely restriction element could not be singled out in this case.

The most frequent class II HLA molecule present in TcTSp5-reactive patients was HLA-DPA1*01:03-DPB1*04:02 (86% of the positive patients). However, this allele had also a high occurrence in TcTSp5-nonresponding patients (52%), and the peptide was predicted to bind this molecule only very weakly (prediction percentile ligand likelihood rank score, from this point onwards prediction rank 13.0). This was also the case for the variant with the largest frequency difference between responding and nonresponding patients, HLA-DQA1*02:01-DQB1*02:02 (57% of the responding patients versus 9% of the nonresponding patients; prediction rank 13.0). An alternative candidate restriction element was DRB1*07:01. This molecule presented both high frequency in responding patients and frequency difference between responding and nonresponding patients and was predicted to be a strong binder of peptide TcTSp5 (prediction rank 1.7). We therefore suggest DRB1*07:01 as the most likely restriction allele for the epitope in peptide TcTSp5.

Regarding TcCZp2, the only class I HLA allele shared by all three IFN-γ–positive patients was HLA-A*31:01. HLA-A*31:01 was predicted to bind the peptide RYQPYHSRHR contained within TcCZp2 with a prediction rank of 0.14 and affinity of 19.2 nM; also, there were two other nested peptides within the 15-mer predicted to bind to HLA-A*31:01 with prediction rank values <0.50. Hence, this allele was determined to be the most likely restriction allele for the epitope(s) contained in peptide TcCZp2.

To confirm the HLA-restricted epitope presentation and to further characterize the phenotype of epitope-specific T cells, fluorescent-labeled peptide–HLA multimer molecules were synthesized for peptide TcTSp5 bound to HLA-DRB1*07:01 and the predicted nested core RYQPYHSRHR of peptide TcCZp2, from this point onwards TcCZp2me, bound to HLA-A*31:01. These were used, along with a panel of T cell surface markers, to determine the differentiation profile of epitope-specific cells by flow cytometry. Fig. 3A shows the gating procedure used for these analyses.

FIGURE 3.

Flow cytometric analysis of epitope-specific T cells. (A) Representative gating strategy implemented using the DAFi method for the analysis of MHC–peptide tetramer–stained cells. (B) Frequency of HLA-A*31:01 (TcCzp2me) tetramer–positive CD8+ T cells found in PBMC from HLA-A*31:01+ patients, ex vivo (top panel) and after 1 wk of in vitro culture with or without stimulation with TcCzp2 or TcCzp2me. (C) Subset distribution of multimer-positive CD8+ T cells in the same experiment as (B), according to their expression of markers CD45RA and CCR7. The scale of red color intensity represents the frequency of tetramer-positive cells corresponding to each patient and condition. Donor identifiers (IDs) beginning with A and C indicate donors classified as asymptomatic and cardiac patients, respectively. TN, naive T cell.

FIGURE 3.

Flow cytometric analysis of epitope-specific T cells. (A) Representative gating strategy implemented using the DAFi method for the analysis of MHC–peptide tetramer–stained cells. (B) Frequency of HLA-A*31:01 (TcCzp2me) tetramer–positive CD8+ T cells found in PBMC from HLA-A*31:01+ patients, ex vivo (top panel) and after 1 wk of in vitro culture with or without stimulation with TcCzp2 or TcCzp2me. (C) Subset distribution of multimer-positive CD8+ T cells in the same experiment as (B), according to their expression of markers CD45RA and CCR7. The scale of red color intensity represents the frequency of tetramer-positive cells corresponding to each patient and condition. Donor identifiers (IDs) beginning with A and C indicate donors classified as asymptomatic and cardiac patients, respectively. TN, naive T cell.

Close modal

As shown in Fig. 3B, multimer-stained, epitope-specific CD8+ T cells were observed ex vivo in patients C16, A20, and A24, whereas the rest of the HLA-A*31:01+ patients in the cohort did not have a detectable frequency of these cells, at least under this experimental setup. This is in agreement with the results observed by IFN-γ ELISPOT for these donors (Fig. 2).

To assess whether specific cells were present in these subjects but were too few to be detected by this technique, we attempted to expand them by incubating PBMC with either the full-length TcCZp2 or TcCZp2me. Under these conditions, patient C01, for whom epitope-specific cells were not detectable ex vivo, exhibited measurable frequencies of multimer-stained cells (Fig. 3B). Regarding patients C16 and A20, the population of epitope-specific cells was reduced after culture, and this reduction was of a greater magnitude upon incubation with the peptide than in the unstimulated condition. Conversely, PBMC sample from patient A24 maintained a similar proportion of epitope-specific cells after exposure to the peptide than the one found ex vivo, but this population shrunk in vitro when cells were left unstimulated. Absence of an epitope-specific T cell population was observed, even after the in vitro stimulation, in the remaining HLA-A31:01+ patients.

Furthermore, given the extensively reported relationship between the maturation/differentiation profile of memory T cells and their functional profile (3941), we decided to assess the expression of markers CD45RA and CCR7 on the surface of TcCZp2me-specific T cells (Fig. 3C). Epitope-specific cells from patient A24 were observed to present a predominant effector memory re-expressing CD45RA T (TEMRA) cell phenotype [CD45RA+CCR7, (40)] ex vivo, whereas they were predominantly effector memory T (TEM) cells (CD45RACCR7) in patients A20 and C16. This profile was markedly altered in specific cells after in vitro stimulation. A24 cells remained predominantly within the TEMRA cell population when they were stimulated with the full TcCZp2 but switched to a TEM phenotype when stimulated with TcCZp2me or when they were left unstimulated. Conversely, the frequency of specific TEMRA cells from patient C16 increased, whereas that of TEM decreased, in the absence of the peptides or upon stimulation with core peptide TcCZp2me, but they retained their distribution among these profiles when they were incubated with TcCZp2. In both cases, a slight increase in the representation of central memory T (TCM; CD45RACCR7+) cells was also observed. Finally, the epitope-specific cells arising from in vitro peptide stimulation of PBMC from patient C01 were predominantly TCM cells upon incubation with either the full length or the core peptide.

Unfortunately, DRB1*07:01–TcTSp5 complexes produced equal levels of staining in CD4+ and CD8+ T cells, suggesting that the resulting multimer staining was not specific enough. Hence, the outcome of these experiments was deemed unreliable, and we were, therefore, not able to use this technology to confirm the HLA restriction of TcTSp5. However, the fact that the multimer–peptide molecular complex was successfully produced and folded, according to the manufacturer-provided quality control data, is a strong indicator that TcTSp5 is indeed able to bind the DRB1*07:01 HLA variant.

Next, IFN-γ intracellular cytokine staining was performed on peptide-stimulated PBMC as an approach to the study of epitope-specific cells in the cases of CD4+ activating peptides (Fig. 4). The sensitivity of this method restrained the detection of cytokine-producing cells to the peptide–patient pairs with the highest SFU counts in the prior ELISPOT experimental setup. As shown in Fig. 4A and 4B, upon stimulation with peptide TcTSp4, IFN-γ production was observed in CD4+ T cells from patients A03 (0.42% of total T cells and 0.50% of CD4+ T cells) and A11 (0.21% of total T cells and 0.26% of CD4+ T cells), and these cells were predominantly TEM and TCM cells (Fig. 4C). The latter patient also exhibited IFN-γ+ cells within the CD8+ T cell subset (0.09% of total T cells and 0.48% of CD8+ T cells), which is in line with the results observed in the depletion ELISPOT experiments from Fig. 2. In contrast, IFN-γ–producing CD4+ T cells were also detected in response to peptide TcTSp5 in PBMC from patient A05 (0.04% of total T cells and 0.07% of CD4+ T cells), but these were predominantly TEM and TEMRA cells.

FIGURE 4.

IFN-γ intracellular cytokine staining. Gating strategy was identical to that used for the experiments depicted on Fig. 3, with the exception of the staining with BV421-labeled anti–IFN-γ Ab instead of BV421-labeled HLA–peptide multimers. (A) Representative cytograms from one out of three patients with measurable IFN-γ+ T cell populations. The represented events were gated on live CD3+ cells. (B) Frequency of IFN-γ+ events within the CD3+ subset, segregated according to their expression of CD4 or CD8. The peptide used to challenge each patients’ PBMC is indicated below the patient identifier (ID) in brackets. (C) Subset distribution of IFN-γ+CD4+ T cells in the same experiment as (B), according to their expression of markers CD45RA and CCR7. Donor IDs beginning with A indicate donors classified as asymptomatic. TN, naive T cell.

FIGURE 4.

IFN-γ intracellular cytokine staining. Gating strategy was identical to that used for the experiments depicted on Fig. 3, with the exception of the staining with BV421-labeled anti–IFN-γ Ab instead of BV421-labeled HLA–peptide multimers. (A) Representative cytograms from one out of three patients with measurable IFN-γ+ T cell populations. The represented events were gated on live CD3+ cells. (B) Frequency of IFN-γ+ events within the CD3+ subset, segregated according to their expression of CD4 or CD8. The peptide used to challenge each patients’ PBMC is indicated below the patient identifier (ID) in brackets. (C) Subset distribution of IFN-γ+CD4+ T cells in the same experiment as (B), according to their expression of markers CD45RA and CCR7. Donor IDs beginning with A indicate donors classified as asymptomatic. TN, naive T cell.

Close modal

The identification of T cell epitopes is of central interest for the understanding of specific adaptive immunity to infections and for the ulterior development of prophylactic and therapeutic immune-mediated interventions. In this study, by means of predictive tools and immunoassaying, we report, to our knowledge, the discovery of novel T. cruzi epitopes and their ability to activate memory CD4+ and CD8+ T cells from chronic Chagas’ disease patients.

The initial universe of sequences used as input for our pipeline was restricted to T. cruzi proteins with positive T cell assays, in either human or nonhuman experimental models, rendering a set of sequences small enough to be addressable by the pipeline while still providing a broad scope of analysis.

The second input required by the strategy hereby presented is a set of HLA variants, coding for molecules against which peptide binding is to be predicted. For lack of pertinent Chagas’ disease– or endemic area–specific information, we selected HLA-A and -B variants based on prevalence in Latin American population and HLA-DRB [note that the HLA-DRA gene is essentially monomorphic (42)] variants based on worldwide prevalence data. In that regard, directed HLA-typing studies would be significantly beneficial for our approach and others’ (43, 44). Nonetheless, a large proportion of the HLA variants later found in our population sample were included in the initial selection, especially within the HLA-A locus (Supplemental Fig. 1A, 1C). Moreover, all patients in the cohort carry at least one of the alleles in the set used for the bioinformatic prediction (Supplemental Fig. 1B).

Given their lack of clinical manifestations of the disease, it has been suggested that asymptomatic chronic Chagas patients have a better immune control of T. cruzi infection than cardiac patients (10, 16). Besides, as mentioned above, T cells have been proven important for both host defense and pathogenesis in Chagas’ disease (16). Hence, we were prompted to consider the clinical status of the donors in the analysis of the response against our candidate epitopes. However, no relationship was found in our data between chronic Chagas cardiopathy and the number of epitopes recognized, the frequency of responding patients, the frequency of IFN-γ–producing cells per patient, or the phenotype of peptide-responsive cells.

It should be noted that this is not the first study presenting a predictive approach applied to the study of T cell specificity in the context of chronic Chagas’ disease. Martin and collaborators (14) used sequence homology to identify raw T. cruzi DNA–sequencing reads matching known HLA-A*02:01–binding trans-sialidase peptides, which, at their time, had been discovered by conformity with the expanded HLA-A*02:01 binding motif (45). Fonseca and collaborators (13) used a similar, motif-based approach in combination with neural network–powered algorithms (46) to predict HLA-A*02:01–binding peptides derived from the sequences of FL-160 (another trans-sialidase) and cruzipain. Targeting the same HLA variant, class I epitopes recognized by patient cells were identified in proteins HSP-70 (47) and TcCA-2 (15) by means of the predictive algorithm SYPFEITHI. In addition to departing from a larger set of proteins, there are other features that set our approach apart from the aforementioned: first, the predictive algorithm NetMHCpan (26, 48, 49), based in artificial neural networks trained with empirical HLA binding and immunopeptidomics data, is regarded as a state-of-the-art method for class I epitope prediction (50). Second, the use of NetMHCpan and NetMHCIIpan (27, 51, 52) enabled the prediction of peptide binding, not only to multiple class I HLA alleles, but also to class II HLA alleles. Finally, the selection was conducted spanning a large set of HLA alleles using the PopCover method, which allowed the selection of a peptide set with broad both allelic and antigenic coverage.

In relation to this, peptides TcTSp1, TcTSp2, and TcCZp1 from our study comprised subsequences previously studied as CD8+ T cell epitopes (13, 14). In these preceding reports, the peptide FANYKFTLV contained in TcTSp1 was found to induce IFN-γ secretion in cells from some, but not all, HLA-A*02:01+ Chagas’ disease patients (14). Meanwhile, truncated versions of peptides TcTSp2 and TcCZp1 were deemed uncapable of inducing IFN-γ secretion in patient PBMC samples (13, 14), even though some of these sequences were proven to bind HLA-A*02:01 in vitro (13). Our in silico analysis predicted that, out of TcTSp1, TcTSp2, and TcCZp1, only TcCZp1 may bind strongly enough to HLA-A*02:01 to generate a memory T cell response. However, the results obtained in this study indicate that, at least among the analyzed patients, IFN-γ response against this peptide was predominantly due to CD4+ T cell activation. Moreover, the only HLA molecule expressed by all three patients whose samples showed response against TcTSp1 was that encoded by the allele combination HLA-DPA1*01:03-DPB1*04:02. This peptide–HLA combination resulted positive for predicted binding (prediction rank, 0.7; affinity, 14.94 nM). Regarding peptide TcTSp2, patients with IFN-γ response against it share two HLA variants predicted as binders: HLA-DRB1*07:01 (prediction rank, 2.0; affinity, 35.65), and HLA-DPA1*01:03-DPB1*04:02 (prediction rank, 1.4; affinity, 21.59 nM). Thus, despite their embedded, previously reported class I candidate epitopes, TcTSp1, TcTSp2, and TcCZp1 also constitute or contain CD4+ T cell epitopes from T. cruzi that, to the best of our knowledge, had not been reported.

The description of new class II epitopes is a key contribution of this work, as the study of these cells’ specificity in Chagas’ disease is far less documented than that of CD8+ T cells. Until the production of this article, T. cruzi proteins known to activate a Th response in Chagas’ disease patients were KMP-11 (17), Tc24, and TSA-1 (18). Abel and collaborators (19) had also determined that several variants of a peptide from repetitive Ag B13 are presented to CD4+ T cells in the context of HLA-DQA1*05:01-DQB1*03:01, and some of these variants trigger proliferation specifically in cells from T. cruzi–infected patients. To our knowledge, in this study, we first report the activation of Chagas’ disease patient–derived CD4+ T cells by epitopes from several putative trans-sialidases (peptides TcTSp1, TcTSp3, and TcTSp4), FL-160 (TcTSp2), cruzipain (TcCZp1), and ASP-2 (TcTSp5).

We also introduce a new class I T. cruzi epitope, denoted TcCZp2me (RYQPYHSRHR), from cruzipain, a known CD8+ T cell–activating Ag (13). This peptide was confirmed to be presented in the context of HLA-A*31:01. Epitope-specific cells were detected ex vivo in three out of eight analyzed patients who express this HLA variant, in frequencies between 0.13 and 0.23% of total CD8+ T cells. The phenotype of these cells was consistent with a high degree of differentiation (TEM and TEMRA cells). Of note, CD8+ TEM and TEMRA cells have been demonstrated to constitute a memory population with preserved proliferative capacity and rapid effector response and to be major contributors to recall response in the context of different acute and chronic viral infections (5357) or vaccination with live attenuated virus (58). In particular, TEMRA cells have been reported to express a transcriptional program that is significantly different from that of TEM cells, which suggests a more specialized functional profile (55). In addition, their frequency has been correlated to enhanced control of certain viral infections (54, 57). Interestingly, the reactivation of CD45RA expression on Ag-experienced, epitope-specific CD8+ T cells requires cytokine-driven proliferation in the absence of TCR stimulation (59, 60). Thus, predominance of TEM epitope–specific cells may suggest the presence of an active site of antiparasitic cytotoxic response at the time the sample was collected. Similarly, TEMRA cells predominance among epitope-specific cells may be a proxy indicator of a “controlled” infection focus, in which the Ag source has been neutralized, and the proliferation of such specific clones has been induced by the action of cytokines in the absence of TCR engagement. However, testing this hypothesis will certainly require further experimentation and escapes the scope of this report.

Data obtained after the stimulation of PBMC from HLA-A*31:01+ subjects with the full and core versions of TcCZp2 showed dissimilar behaviors in terms of the resulting frequency of specific cells and their phenotype. For subjects C16 and A20, the epitope-specific population seemed to shrink faster upon stimulation than if left unstimulated. Because, for these two donors, epitope-specific cells were predominantly TEM and TEMRA cells, this observation is in line with the model proposed by Geginat and collaborators (59), by which IL-15 induces a prompter proliferative response on CD8+ TEM and TEMRA cells than it does on their TCM counterparts, whereas the former subsets are more susceptible to apoptosis upon engagement of the TCR. Our results also agree with this model in that these epitope-specific cells gained expression of CD45RA when they were cultured in the absence of their cognate peptide. Additionally, activation-induced TCR endocytosis may have also contributed to the lower detection of epitope-specific cells after in vitro stimulation (61, 62). In fact, decreased tetramer mean fluorescence intensity was observed in tetramer-positive cells after incubation with TcCZp2me, but not TcCZp2, compared with those in the unstimulated condition (data not shown).

In contrast, TcCZp2me-specific cells from patient A24 reduced their frequency when they were cultured without stimulation, but were preserved upon addition of the full length or the core versions of the peptide. Noticeably, the phenotype of these cells after stimulation was different depending on whether they had been exposed to TcCZp2 or TcCZp2me. This may indicate that each of these peptides triggered preferentially bystander or TCR-mediated activation of TcCZp2me-specific cells upon stimulation, respectively. In the case of patient C01, for whom TcCZp2me-specific cells were undetectable ex vivo but were expanded after peptide stimulation, the preponderance of a TCM phenotype in contrast with the effector phenotype observed in the previous cases is suggestive of their origin as in vitro activated naive T cells (63). Overall, our phenotyping data clearly show that epitope-specific CD8+ T cells have a heterogeneous degree of maturation across the chronic Chagas’ disease patient population. This holds true for the CD4+ T cells specific for TcTSp4 and TcTSp5, although TCM cells seemed to be preponderant in patients in whom IFN-γ response against TcTSp4 was detected by intracellular cytokine staining.

There are a number of factors that determine the acquisition of a memory T cell population with a certain specificity (6467). Several of these, like the expression of costimulatory molecules on the surface of APC or the secretion of certain cytokines, have been shown to be altered or subverted by T. cruzi (16). Thus, the generation of parasite-specific memory T cells may, in consequence, be affected by the infection itself. As a result, the expression of an HLA molecule capable of presenting a certain T. cruzi peptide does not guarantee that a memory T cell recall response against that peptide will be observed in the circulation of a patient, at least in the chronic phase of infection. This is illustrated by the data presented in this study: out of 13 HLA-A*31:01+ chronic Chagas patients for whom IFN-γ secretion was assessed against TcCZp2me and out of seven for whom the presence of specific cells was directly measured ex vivo, only three resulted positive. Likewise, cells producing IFN-γ upon stimulation with peptide TcTSp5 were found in four out of nine HLA-DRB1*07:01+ patients in the cohort. The apparent in vitro activation of TcCZp2me-specific naive T cells in PBMC from patient C01 further supports this remark. Hence, HLA restriction does not fully explain the presence or absence of response toward a certain epitope. It could be fairly argued that using a single cytokine as readout poses the risk of missing the presence of specific cells with a different response profile (31). However, only the subjects for whom IFN-γ secretion was observed against peptide TcCZp2 exhibited a detectable frequency of HLA multimer–stained cells ex vivo. In addition, samples from HLA-A*31:01+ and HLA-DRB1*07:01+ patients were challenged with TcCZp2 and TcTSp5, respectively, in IL-10 ELISPOT experiments, with results showing no difference upon peptide stimulation from the nonstimulated control conditions (data not shown).

The intraspecies phylogenetic diversity of T. cruzi is often regarded as a source of variability in parameters related to Chagas’ disease epidemiology and pathophysiology (68). In direct relation to this work, the repertoire of epitopes presented to T cells in the context of infection is very likely to also be subjected to such variability. To tackle this issue, we only included sequences conserved in the genomes of the divergent T. cruzi strains Sylvio X10, representative of discrete typing unit (DTU) TcI, and CL Brener, representative of DTU TcVI. Nonetheless, the whole spectrum of sequence variability in the proteomes of natural populations of the parasite, even within a single DTU, can hardly be encompassed by only two strains. In particular, the trans-sialidase family of proteins is outstandingly expanded in T. cruzi (69), which positions it as a source of a large number of variation-prone epitopes.

Regarding their immune relevance at the human population level, each of the epitopes described in this article is recognized, at most, by one fifth of the Chagas’ disease patients in our cohort. The combination of all seven peptides reached a maximum coverage of 37.3% of the population sample. Taken together, this and the previously discussed results suggest that a T cell–based vaccine or immune intervention to prevent or treat Chagas’ disease will most likely require multiple epitopes to achieve a broad population coverage.

Despite the limited size of our initial universe of peptides, we were able to pinpoint seven peptidic sequences that are or contain previously unknown T. cruzi epitopes. To enhance the reach of this strategy, deeper knowledge is required on which parasite proteins can be efficiently processed and presented in the context of infection. Also, it would be helpful to assess whether the peptides thus generated have noncanonical length and/or anchoring to the HLA molecules [as it was demonstrated to occur in cells infected by other protozoan parasites (70)] or if there is a significant contribution of posttranslational peptide splicing (7173) to the universe of T. cruzi T cell epitopes. We are currently undertaking the exploration of the T. cruzi immunopeptidome to produce a broader, information-based selection of proteins onto which the approach presented in this study could be applied. Nonetheless, as a proof of concept, this study demonstrates that the developed pipeline can be successfully used to identify T. cruzi class I and II epitopes recognized by T cells from chronically infected patients.

We thank the sample donors involved in this study for kind contributions. Also, we thank Drs. Violeta Chiauzzi, Susana Laucella, and Melisa Castro Eiro for providing disinterested technical assistance; Paula Beati for helpful guidance on bioinformatics and programming; and Dr. Silvia Longhi for critically revising the manuscript.

This work was supported by funds granted by Argentina’s National Agency for Scientific and Technological Promotion to K.A.G. (PICT 2014-1026) and to M.N. (PICT 2016-0089), and by the Alberto J. Roemmers Foundation to G.R.A. (Research Grant for Novel Investigators 2016–2018).

The online version of this article contains supplemental material.

Abbreviations used in this article:

     
  • DAFi

    directed automated filtering and identification of cell populations

  •  
  • DTU

    discrete typing unit

  •  
  • KMP-11

    kinetoplastid membrane protein 11

  •  
  • SFU

    spot-forming unit

  •  
  • TCM

    central memory T

  •  
  • TEM

    effector memory T

  •  
  • TEMRA

    effector memory re-expressing CD45RA T.

1
Sartori
,
A. M. C.
,
J. E.
Neto
,
E. V.
Nunes
,
L. M. A.
Braz
,
H. H.
Caiaffa-Filho
,
O.
da Cruz Oliveira
Jr.
,
V. A.
Neto
,
M. A.
Shikanai-Yasuda
.
2002
.
Trypanosoma cruzi parasitemia in chronic Chagas disease: comparison between human immunodeficiency virus (HIV)-positive and HIV-negative patients.
J. Infect. Dis.
186
:
872
875
.
2
Vaidian
,
A. K.
,
L. M.
Weiss
,
H. B.
Tanowitz
.
2004
.
Chagas’ disease and AIDS.
Kinetoplastid Biol. Dis.
3
:
2
.
3
Dos Santos Virgilio
,
F.
,
C.
Pontes
,
M. R.
Dominguez
,
J.
Ersching
,
M. M.
Rodrigues
,
J. R.
Vasconcelos
.
2014
.
CD8+ T cell-mediated immunity during Trypanosoma cruzi infection: a path for vaccine development?
Mediators Inflamm.
2014
:
243786
.
4
Engman
,
D. M.
,
J. S.
Leon
.
2002
.
Pathogenesis of Chagas heart disease: role of autoimmunity.
Acta Trop.
81
:
123
132
.
5
Scharfstein
,
J.
,
J.
de Assis Silva Gomes
,
R.
Correa-Oliveira
.
2009
.
Back to the future in Chagas disease: from animal models to patient cohort studies, progress in immunopathogenesis research.
Mem. Inst. Oswaldo Cruz
104
(
Suppl. 1
):
187
198
.
6
Bonney
,
K. M.
,
D. M.
Engman
.
2015
.
Autoimmune pathogenesis of Chagas heart disease: looking back, looking ahead.
Am. J. Pathol.
185
:
1537
1547
.
7
Lannes-Vieira
,
J.
2003
.
Trypanosoma cruzi-elicited CD8+ T cell-mediated myocarditis: chemokine receptors and adhesion molecules as potential therapeutic targets to control chronic inflammation?
Mem. Inst. Oswaldo Cruz
98
:
299
304
.
8
da Silveira
,
A. B. M.
,
E. M.
Lemos
,
S. J.
Adad
,
R.
Correa-Oliveira
,
J. B.
Furness
,
D.
D’Avila Reis
.
2007
.
Megacolon in Chagas disease: a study of inflammatory cells, enteric nerves, and glial cells.
Hum. Pathol.
38
:
1256
1264
.
9
Fonseca
,
S. G.
,
M. M.
Reis
,
V.
Coelho
,
L. G.
Nogueira
,
S. M.
Monteiro
,
E. C.
Mairena
,
F.
Bacal
,
E.
Bocchi
,
L.
Guilherme
,
X. X.
Zheng
, et al
.
2007
.
Locally produced survival cytokines IL-15 and IL-7 may be associated to the predominance of CD8+ T cells at heart lesions of human chronic Chagas disease cardiomyopathy.
Scand. J. Immunol.
66
:
362
371
.
10
Dutra
,
W. O.
,
K. J.
Gollob
.
2008
.
Current concepts in immunoregulation and pathology of human Chagas disease.
Curr. Opin. Infect. Dis.
21
:
287
292
.
11
Wizel
,
B.
,
M.
Nunes
,
R. L.
Tarleton
.
1997
.
Identification of Trypanosoma cruzi trans-sialidase family members as targets of protective CD8+ TC1 responses.
J. Immunol.
159
:
6120
6130
.
12
Low
,
H. P.
,
M. A.
Santos
,
B.
Wizel
,
R. L.
Tarleton
.
1998
.
Amastigote surface proteins of Trypanosoma cruzi are targets for CD8+ CTL.
J. Immunol.
160
:
1817
1823
.
13
Fonseca
,
S. G.
,
H.
Moins-Teisserenc
,
E.
Clave
,
B.
Ianni
,
V. L.
Nunes
,
C.
Mady
,
L. K.
Iwai
,
A.
Sette
,
J.
Sidney
,
M. L. C.
Marin
, et al
.
2005
.
Identification of multiple HLA-A*0201-restricted cruzipain and FL-160 CD8+ epitopes recognized by T cells from chronically Trypanosoma cruzi-infected patients.
Microbes Infect.
7
:
688
697
.
14
Martin
,
D. L.
,
D. B.
Weatherly
,
S. A.
Laucella
,
M. A.
Cabinian
,
M. T.
Crim
,
S.
Sullivan
,
M.
Heiges
,
S. H.
Craven
,
C. S.
Rosenberg
,
M. H.
Collins
, et al
.
2006
.
CD8+ T-cell responses to Trypanosoma cruzi are highly focused on strain-variant trans-sialidase epitopes.
PLoS Pathog.
2
:
e77
.
15
Egui
,
A.
,
M. C.
Thomas
,
B.
Carrilero
,
M.
Segovia
,
C.
Alonso
,
C.
Marañón
,
M. C.
López
.
2015
.
Differential phenotypic and functional profiles of TcCA-2 -specific cytotoxic CD8+ T cells in the asymptomatic versus cardiac phase in Chagasic patients.
PLoS One
10
:
e0122115
.
16
Acevedo
,
G. R.
,
M. C.
Girard
,
K. A.
Gómez
.
2018
.
The unsolved jigsaw puzzle of the immune response in Chagas disease.
Front. Immunol.
9
:
1929
.
17
Cuellar
,
A.
,
F.
Rojas
,
N.
Bolaños
,
H.
Diez
,
M.
Del Carmen Thomas
,
F.
Rosas
,
V.
Velasco
,
M. C.
López
,
J. M.
González
,
C.
Puerta
.
2009
.
Natural CD4+ T-cell responses against Trypanosoma cruzi KMP-11 protein in chronic chagasic patients. [Published erratum appears in 2009 Immunol. Cell Biol. 87: 108.]
Immunol. Cell Biol.
87
:
149
153
.
18
Villanueva-Lizama
,
L. E.
,
J. V.
Cruz-Chan
,
A. D. C.
Aguilar-Cetina
,
L. F.
Herrera-Sanchez
,
J. M.
Rodriguez-Perez
,
M. E.
Rosado-Vallado
,
M. J.
Ramirez-Sierra
,
J.
Ortega-Lopez
,
K.
Jones
,
P.
Hotez
, et al
.
2018
.
Trypanosoma cruzi vaccine candidate antigens Tc24 and TSA-1 recall memory immune response associated with HLA-A and -B supertypes in Chagasic chronic patients from Mexico.
PLoS Negl. Trop. Dis.
12
:
e0006240
.
19
Abel
,
L. C. J.
,
L. K.
Iwai
,
W.
Viviani
,
A. M.
Bilate
,
K. C.
Faé
,
R. C.
Ferreira
,
A. C.
Goldberg
,
L.
Juliano
,
M. A.
Juliano
,
B.
Ianni
, et al
.
2005
.
T cell epitope characterization in tandemly repetitive Trypanosoma cruzi B13 protein.
Microbes Infect.
7
:
1184
1195
.
20
Hondowicz
,
B. D.
,
K. V.
Schwedhelm
,
A.
Kas
,
M. A.
Tasch
,
C.
Rawlings
,
N.
Ramchurren
,
M.
McIntosh
,
L. A.
D’Amico
,
S.
Sanda
,
N. E.
Standifer
, et al
.
2012
.
Discovery of T cell antigens by high-throughput screening of synthetic minigene libraries.
PLoS One
7
:
e29949
.
21
El-Sayed
,
N. M.
,
P. J.
Myler
,
D. C.
Bartholomeu
,
D.
Nilsson
,
G.
Aggarwal
,
A.-N.
Tran
,
E.
Ghedin
,
E. A.
Worthey
,
A. L.
Delcher
,
G.
Blandin
, et al
.
2005
.
The genome sequence of Trypanosoma cruzi, etiologic agent of Chagas disease.
Science
309
:
409
415
.
22
Berná
,
L.
,
M.
Rodriguez
,
M. L.
Chiribao
,
A.
Parodi-Talice
,
S.
Pita
,
G.
Rijo
,
F.
Alvarez-Valin
,
C.
Robello
.
2018
.
Expanding an expanded genome: long-read sequencing of Trypanosoma cruzi.
Microb. Genom.
DOI: 0.1099/mgen.0.000177.
23
Grubaugh
,
D.
,
J. B.
Flechtner
,
D. E.
Higgins
.
2013
.
Proteins as T cell antigens: methods for high-throughput identification.
Vaccine
31
:
3805
3810
.
24
National Institute of Allergy and Infectious Diseases
. Immune Epitope Database. Available at: http://www.iedb.org/. Accessed: August 7,
2013
.
25
The EuPathDB Project Team
. TriTrypDB Database, Version 6.0. Available at: http://www.tritrypdb.org. Accessed: between September 25, 2013 and January 28, 2014.
26
Nielsen
,
M.
,
C.
Lundegaard
,
T.
Blicher
,
K.
Lamberth
,
M.
Harndahl
,
S.
Justesen
,
G.
Røder
,
B.
Peters
,
A.
Sette
,
O.
Lund
,
S.
Buus
.
2007
.
NetMHCpan, a method for quantitative predictions of peptide binding to any HLA-A and -B locus protein of known sequence.
PLoS One
2
:
e796
.
27
Nielsen
,
M.
,
S.
Justesen
,
O.
Lund
,
C.
Lundegaard
,
S.
Buus
.
2010
.
NetMHCIIpan-2.0 - improved pan-specific HLA-DR predictions using a novel concurrent alignment and weight optimization training procedure.
Immunome Res.
6
:
9
.
28
Lundegaard
,
C.
,
C.
Perez
.
2010
. PopCover – a method for selecting of peptides with optimal population and pathogen coverage. In Proceedings of the 1st ACM International Conference on Bioinformatics and Computational Biology, August 2–4, 2010,
Association for Computing Machinery
, New York, NY, p.
658
659
.
29
Longhi
,
S. A.
,
A.
Atienza
,
G.
Perez Prados
,
A.
Buying
,
V.
Balouz
,
C. A.
Buscaglia
,
R.
Santos
,
L. M.
Tasso
,
R.
Bonato
,
P.
Chiale
, et al
.
2014
.
Cytokine production but lack of proliferation in peripheral blood mononuclear cells from chronic Chagas’ disease cardiomyopathy patients in response to T. cruzi ribosomal P proteins.
PLoS Negl. Trop. Dis.
DOI: 10.1371/journal.pntd.0002906.
30
Wittig
,
M.
,
J. A.
Anmarkrud
,
J. C.
Kässens
,
S.
Koch
,
M.
Forster
,
E.
Ellinghaus
,
J. R.
Hov
,
S.
Sauer
,
M.
Schimmler
,
M.
Ziemann
, et al
.
2015
.
Development of a high-resolution NGS-based HLA-typing and analysis pipeline.
Nucleic Acids Res.
43
:
e70
.
31
Acevedo
,
G. R.
,
S. A.
Longhi
,
A.
Bunying
,
N.
Sabri
,
A.
Atienza
,
M. P.
Zago
,
R.
Santos
,
V. A.
Judkowski
,
C.
Pinilla
,
K. A.
Gómez
.
2017
.
Methodological approach to the ex vivo expansion and detection of T. cruzi-specific T cells from chronic Chagas disease patients. [Published erratum appears in 2017 PLoS One 12: e0184467.]
PLoS One
12
:
e0178380
.
32
Finak
,
G.
,
J.
Frelinger
,
W.
Jiang
,
E. W.
Newell
,
J.
Ramey
,
M. M.
Davis
,
S. A.
Kalams
,
S. C.
De Rosa
,
R.
Gottardo
.
2014
.
OpenCyto: an open source infrastructure for scalable, robust, reproducible, and automated, end-to-end flow cytometry data analysis.
PLOS Comput. Biol.
10
:
e1003806
.
33
Lee
,
A. J.
,
I.
Chang
,
J. G.
Burel
,
C. S.
Lindestam Arlehamn
,
A.
Mandava
,
D.
Weiskopf
,
B.
Peters
,
A.
Sette
,
R. H.
Scheuermann
,
Y.
Qian
.
2018
.
DAFi: a directed recursive data filtering and clustering approach for improving and interpreting data clustering identification of cell populations from polychromatic flow cytometry data.
Cytometry A
93
:
597
610
.
34
Wickham
,
H.
2016
.
ggplot2: Elegant Graphics for Data Analysis
, 2nd Ed.
Springer Verlag
,
New York
.
35
Moodie
,
Z.
,
L.
Price
,
C.
Gouttefangeas
,
A.
Mander
,
S.
Janetzki
,
M.
Löwer
,
M. J.
Welters
,
C.
Ottensmeier
,
S. H.
van der Burg
,
C. M.
Britten
.
2010
.
Response definition criteria for ELISPOT assays revisited.
Cancer Immunol. Immunother.
59
:
1489
1501
.
36
Tzelepis
,
F.
,
B. C. G.
de Alencar
,
M. L. O.
Penido
,
C.
Claser
,
A. V.
Machado
,
O.
Bruna-Romero
,
R. T.
Gazzinelli
,
M. M.
Rodrigues
.
2008
.
Infection with Trypanosoma cruzi restricts the repertoire of parasite-specific CD8+ T cells leading to immunodominance.
J. Immunol.
180
:
1737
1748
.
37
Bustamante
,
J. M.
,
L. M.
Bixby
,
R. L.
Tarleton
.
2008
.
Drug-induced cure drives conversion to a stable and protective CD8+ T central memory response in chronic Chagas disease.
Nat. Med.
14
:
542
550
.
38
Bixby
,
L. M.
,
R. L.
Tarleton
.
2008
.
Stable CD8+ T cell memory during persistent Trypanosoma cruzi infection.
J. Immunol.
181
:
2644
2650
.
39
Zielinski
,
C. E.
,
D.
Corti
,
F.
Mele
,
D.
Pinto
,
A.
Lanzavecchia
,
F.
Sallusto
.
2011
.
Dissecting the human immunologic memory for pathogens.
Immunol. Rev.
240
:
40
51
.
40
Sallusto
,
F.
,
J.
Geginat
,
A.
Lanzavecchia
.
2004
.
Central memory and effector memory T cell subsets: function, generation, and maintenance.
Annu. Rev. Immunol.
22
:
745
763
.
41
Busch
,
D. H.
,
S. P.
Fräßle
,
D.
Sommermeyer
,
V. R.
Buchholz
,
S. R.
Riddell
.
2016
.
Role of memory T cell subsets for adoptive immunotherapy.
Semin. Immunol.
28
:
28
34
.
42
Stern
,
L. J.
,
J. M.
Calvo-Calle
.
2009
.
HLA-DR: molecular insights and vaccine design.
Curr. Pharm. Des.
15
:
3249
3261
.
43
Bustamante
,
C. D.
,
E. G.
Burchard
,
F. M.
De la Vega
.
2011
.
Genomics for the world.
Nature
475
:
163
165
.
44
Jooma
,
S.
,
M. J.
Hahn
,
L. A.
Hindorff
,
V. L.
Bonham
.
2019
.
Defining and achieving health equity in genomic medicine.
Ethn. Dis.
29
(
Suppl. 1
):
173
178
.
45
Wizel
,
B.
,
M.
Palmieri
,
C.
Mendoza
,
B.
Arana
,
J.
Sidney
,
A.
Sette
,
R.
Tarleton
.
1998
.
Human infection with Trypanosoma cruzi induces parasite antigen-specific cytotoxic T lymphocyte responses.
J. Clin. Invest.
102
:
1062
1071
.
46
Gulukota
,
K.
,
J.
Sidney
,
A.
Sette
,
C.
DeLisi
.
1997
.
Two complementary methods for predicting peptides binding major histocompatibility complex molecules.
J. Mol. Biol.
267
:
1258
1267
.
47
Marañón
,
C.
,
A.
Egui
,
B.
Carrilero
,
M. C.
Thomas
,
M. J.
Pinazo
,
J.
Gascón
,
M.
Segovia
,
M. C.
López
.
2011
.
Identification of HLA-A∗02:01-restricted CTL epitopes in Trypanosoma cruzi heat shock protein-70 recognized by Chagas disease patients.
Microbes Infect.
13
:
1025
1032
.
48
Hoof
,
I.
,
B.
Peters
,
J.
Sidney
,
L. E.
Pedersen
,
A.
Sette
,
O.
Lund
,
S.
Buus
,
M.
Nielsen
.
2009
.
NetMHCpan, a method for MHC class I binding prediction beyond humans.
Immunogenetics
61
:
1
13
.
49
Nielsen
,
M.
,
M.
Andreatta
.
2016
.
NetMHCpan-3.0; improved prediction of binding to MHC class I molecules integrating information from multiple receptor and peptide length datasets.
Genome Med.
8
:
33
.
50
Oyarzun
,
P.
,
B.
Kobe
.
2015
.
Computer-aided design of T-cell epitope-based vaccines: addressing population coverage.
Int. J. Immunogenet.
42
:
313
321
.
51
Karosiene
,
E.
,
M.
Rasmussen
,
T.
Blicher
,
O.
Lund
,
S.
Buus
,
M.
Nielsen
.
2013
.
NetMHCIIpan-3.0, a common pan-specific MHC class II prediction method including all three human MHC class II isotypes, HLA-DR, HLA-DP and HLA-DQ.
Immunogenetics
65
:
711
724
.
52
Nielsen
,
M.
,
C.
Lundegaard
,
T.
Blicher
,
B.
Peters
,
A.
Sette
,
S.
Justesen
,
S.
Buus
,
O.
Lund
.
2008
.
Quantitative predictions of peptide binding to any HLA-DR molecule of known sequence: NetMHCIIpan.
PLOS Comput. Biol.
DOI: 10.1371/journal.pcbi.1000107.
53
Dunne
,
P. J.
,
J. M.
Faint
,
N. H.
Gudgeon
,
J. M.
Fletcher
,
F. J.
Plunkett
,
M. V. D.
Soares
,
A. D.
Hislop
,
N. E.
Annels
,
A. B.
Rickinson
,
M.
Salmon
,
A. N.
Akbar
.
2002
.
Epstein-Barr virus-specific CD8+ T cells that re-express CD45RA are apoptosis-resistant memory cells that retain replicative potential.
Blood
100
:
933
940
.
54
Northfield
,
J. W.
,
C. P.
Loo
,
J. D.
Barbour
,
G.
Spotts
,
F. M.
Hecht
,
P.
Klenerman
,
D. F.
Nixon
,
J.
Michaëlsson
.
2007
.
Human immunodeficiency virus type 1 (HIV-1)-specific CD8+ T(EMRA) cells in early infection are linked to control of HIV-1 viremia and predict the subsequent viral load set point.
J. Virol.
81
:
5759
5765
.
55
Tian
,
Y.
,
M.
Babor
,
J.
Lane
,
G.
Seumois
,
S.
Liang
,
N. D. S.
Goonawardhana
,
A. D.
De Silva
,
E. J.
Phillips
,
S. A.
Mallal
,
R.
da Silva Antunes
, et al
.
2019
.
Dengue-specific CD8+ T cell subsets display specialized transcriptomic and TCR profiles.
J. Clin. Invest.
130
:
1727
1741
.
56
Lilleri
,
D.
,
C.
Fornara
,
M. G.
Revello
,
G.
Gerna
.
2008
.
Human cytomegalovirus-specific memory CD8+ and CD4+ T cell differentiation after primary infection.
J. Infect. Dis.
198
:
536
543
.
57
Sridhar
,
S.
,
S.
Begom
,
A.
Bermingham
,
K.
Hoschler
,
W.
Adamson
,
W.
Carman
,
T.
Bean
,
W.
Barclay
,
J. J.
Deeks
,
A.
Lalvani
.
2013
.
Cellular immune correlates of protection against symptomatic pandemic influenza.
Nat. Med.
19
:
1305
1312
.
58
Akondy
,
R. S.
,
N. D.
Monson
,
J. D.
Miller
,
S.
Edupuganti
,
D.
Teuwen
,
H.
Wu
,
F.
Quyyumi
,
S.
Garg
,
J. D.
Altman
,
C.
Del Río
, et al
.
2009
.
The yellow fever virus vaccine induces a broad and polyfunctional human memory CD8+ T cell response.
J. Immunol.
183
:
7919
7930
.
59
Geginat
,
J.
,
A.
Lanzavecchia
,
F.
Sallusto
.
2003
.
Proliferation and differentiation potential of human CD8+ memory T-cell subsets in response to antigen or homeostatic cytokines.
Blood
101
:
4260
4266
.
60
Jameson
,
S. C.
,
D.
Masopust
.
2018
.
Understanding subset diversity in T cell memory.
Immunity
48
:
214
226
.
61
Martínez-Martín
,
N.
,
E.
Fernández-Arenas
,
S.
Cemerski
,
P.
Delgado
,
M.
Turner
,
J.
Heuser
,
D. J.
Irvine
,
B.
Huang
,
X. R.
Bustelo
,
A.
Shaw
,
B.
Alarcón
.
2011
.
T cell receptor internalization from the immunological synapse is mediated by TC21 and RhoG GTPase-dependent phagocytosis.
Immunity
35
:
208
222
.
62
Čemerski
,
S.
,
J.
Das
,
E.
Giurisato
,
M. A.
Markiewicz
,
P. M.
Allen
,
A. K.
Chakraborty
,
A. S.
Shaw
.
2008
.
The balance between T cell receptor signaling and degradation at the center of the immunological synapse is determined by antigen quality.
Immunity
29
:
414
422
.
63
Willinger
,
T.
,
T.
Freeman
,
H.
Hasegawa
,
A. J.
McMichael
,
M. F. C.
Callan
.
2005
.
Molecular signatures distinguish human central memory from effector memory CD8 T cell subsets.
J. Immunol.
175
:
5895
5903
.
64
Sibener
,
L. V.
,
R. A.
Fernandes
,
E. M.
Kolawole
,
C. B.
Carbone
,
F.
Liu
,
D.
McAffee
,
M. E.
Birnbaum
,
X.
Yang
,
L. F.
Su
,
W.
Yu
, et al
.
2018
.
Isolation of a structural mechanism for uncoupling T cell receptor signaling from peptide-MHC binding.
Cell
174
:
672
687.e27
.
65
Jameson
,
S. C.
,
D.
Masopust
.
2009
.
Diversity in T cell memory: an embarrassment of riches.
Immunity
31
:
859
871
.
66
Boyman
,
O.
,
S.
Létourneau
,
C.
Krieg
,
J.
Sprent
.
2009
.
Homeostatic proliferation and survival of naïve and memory T cells.
Eur. J. Immunol.
39
:
2088
2094
.
67
Jaigirdar
,
S. A.
,
M. K. L.
MacLeod
.
2015
.
Development and function of protective and pathologic memory CD4 T cells.
Front. Immunol.
DOI: 10.3389/fimmu.2015.00456.
68
Zingales
,
B.
2018
.
Trypanosoma cruzi genetic diversity: something new for something known about Chagas disease manifestations, serodiagnosis and drug sensitivity.
Acta Trop.
184
:
38
52
.
69
Padilla
,
A. M.
,
J. M.
Bustamante
,
R. L.
Tarleton
.
2009
.
CD8+ T cells in Trypanosoma cruzi infection.
Curr. Opin. Immunol.
21
:
385
390
.
70
McMurtrey
,
C.
,
T.
Trolle
,
T.
Sansom
,
S. G.
Remesh
,
T.
Kaever
,
W.
Bardet
,
K.
Jackson
,
R.
McLeod
,
A.
Sette
,
M.
Nielsen
, et al
.
2016
.
Toxoplasma gondii peptide ligands open the gate of the HLA class I binding groove.
eLife
DOI: 10.7554/eLife.12556.
71
Liepe
,
J.
,
F.
Marino
,
J.
Sidney
,
A.
Jeko
,
D. E.
Bunting
,
A.
Sette
,
P. M.
Kloetzel
,
M. P. H.
Stumpf
,
A. J. R.
Heck
,
M.
Mishto
.
2016
.
A large fraction of HLA class I ligands are proteasome-generated spliced peptides.
Science
354
:
354
358
.
72
Platteel
,
A. C. M.
,
M.
Mishto
,
K.
Textoris-Taube
,
C.
Keller
,
J.
Liepe
,
D. H.
Busch
,
P. M.
Kloetzel
,
A. J. A. M.
Sijts
.
2016
.
CD8+ T cells of Listeria monocytogenes-infected mice recognize both linear and spliced proteasome products.
Eur. J. Immunol.
46
:
1109
1118
.
73
Faridi
,
P.
,
C.
Li
,
S. H.
Ramarathinam
,
J. P.
Vivian
,
P. T.
Illing
,
N. A.
Mifsud
,
R.
Ayala
,
J.
Song
,
L. J.
Gearing
,
P. J.
Hertzog
, et al
.
2018
.
A subset of HLA-I peptides are not genomically templated: evidence for cis- and trans-spliced peptide ligands.
Sci. Immunol.
DOI: 10.1126/sciimmunol.aar3947.
74
Alvarez
,
M. G.
,
M.
Postan
,
D. B.
Weatherly
,
M. C.
Albareda
,
J.
Sidney
,
A.
Sette
,
C.
Olivera
,
A. H.
Armenti
,
R. L.
Tarletona
,
S. A.
Laucella
.
2008
.
HLA class I-T cell epitopes from trans-sialidase proteins reveal functionally distinct subsets of CD8+ T cells in chronic Chagas disease.
PLoS Negl.
Trop. Dis. DOI: 10.1371/journal.pntd.0000288.

The authors have no financial conflicts of interest.